Find attached some remarks. and comments. E.g. I'm not sure if you really want your vector Preamble to be constructed the way you did.
As Andy pointed out, it will not be necessary to initialize a Matrix element by element to zero. Most of the time you will not need to initialize at all. mathcad will create/enlarge a Matrix as needed and fills in zeros itself.
As written the number of cycles of the m-loop is dependend exponentially by the upper value of the L-range. To find out which routine is the most promising candidate for optimizing you will have to use trace, as Andy pointed out.
EDITED: I additionally attached an streamlined version "updated_3") and hope its correct. The different result are mainly (I think) because you did not reset your vector Binary and so got vectors of wrong length. Again, still read the comments in "updated_2".
The main question now is: Why do you need the L-loop and a 2-dimensional matrix? The row for the largest L will always include all the others. The value of L is used only to determine the largest value for m in the inner loop, so the results should be (and are) the same. L is also used as max-digits for the decimal to binary conversion but has no effect there, as the values converted alway have less than L bits in their full representation.
